In 1988, questionnaires were received from 1,400 twin pairs (17% MZM, 23% MZF, 17% DZM, 19% DZF, 24% DZO) aged 11 to 18, registered with the Australian NHMRC Twin Registry. Twins reported independently on themselves and on the perceived behaviour of their parents, siblings and friends. For smoking and for drinking in the previous month, the prevalence was modelled as a logistic function of age, sex, perceived smoking or drinking behaviour of family and friends, and the Adult Eysenck Personality Questionnaire (AEPQ) scales. Strengths of association were: family behaviour, odds ratio (OR) < or = 2; Extraversion and Psychoticism, interquartile OR approximately 1.6; behaviour of friend, OR approximately 3 to 6. Twin associations were represented by odds ratios. For smoking they were 16 in MZ and 7 in DZ same-sex pairs, and 3 in DZO pairs. Although the former is consistent with genetic factors determining adolescent smoking behaviour, the reduced association in DZO pairs and strong association with smoking by friends argue to the contrary. For drinking, twin odds ratios were 11 in MZM, MZF and DZF pairs, and 4 in DZM and DZO pairs, consistent with genetic factors determining alcohol use in male but not female, adolescents. Twin odds ratios were not influenced by adjustment for the AEPQ scales; this does not support the hypothesis that genetic factors which determine personality also determine smoking or drinking behaviour during adolescence [corrected].